The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from Google search results (titles, URLs, and snippets). This creates a surface where malicious instructions embedded in search results could potentially influence the agent's subsequent logic, especially during the scoring and tiering phase.
Ingestion points: Google search snippets and website titles fetched via the scraper.
Boundary markers: None specified in the instructions to delimit untrusted content.
Capability inventory: Local file writing via command line arguments, and command execution via node scripts.
Sanitization: No sanitization or validation of the search result content is mentioned before processing.
[DYNAMIC_EXECUTION]: The skill exposes a customMapFunction parameter which accepts a raw JavaScript function string. This string is intended to be executed within the runtime environment of the scraping service to transform data, which constitutes a dynamic code execution vector.
[EXTERNAL_DOWNLOADS]: The skill depends on the apidojo~google-search-scraper actor hosted on the Apify platform. While this is a vendor-provided resource, it represents an external dependency required for the skill to function.
[COMMAND_EXECUTION]: The documentation provides examples for executing a local script scripts/run_actor.js using node. This script acts as a wrapper for interacting with the external API and handles file operations such as saving results to JSON or CSV formats.

finding-digital-agencies-via-google-search